AI-Powered Intrusion Detection in Network Traffic

Abstract

: As the internet usage is exponentially increasing and with the emerging cyber threats, the conventional rule-based intrusion detection systems (IDS) are limited to identifying new and advanced attacks. The paper features an artificial intelligence-powered Intrusion Detection System (IDS) utilizing the machine learning methods to detect the malicious actions in the network traffic automatically. The system based on the CICIDS datasets, data preprocessing, feature selection, and a Random Forest classifier was used to determine various types of intrusions with a high success rate. The capability of the system to classify intrusions, generate confidence scores and be interpretable with feature importance analysis is evidenced by real-time packet-level simulation. Besides, the model is combined with a notification system that will notify through email or SMS in case of intrusions, so that timely response and mitigation is realized. The suggested framework does not only improve the detection performance, it also brings in scalability, automation, and transparency, which is appropriate in the present day network setup.